Dear Mr. Hurd: We Agree On Something! 30/70 Is The Right Goal For Enterprise IT.

Some customers we share with HP asked about CEO
Mark Hurd’s keynote address at their event in Vegas last week
(I guess
everything that happens in Vegas doesn’t stay in Vegas).

Hurd said the
company needs to reduce costs so it can invest more in innovation, a worthwhile
goal indeed. Hurd said capital spending is flat, but IT maintenance and
operations costs are climbing. He pledged
HP would take maintenance and operations to 30% of the IT budget and innovation
to 70%. We wholeheartedly agree with the goal of 30/70 and, in fact, we’ve made
substantial progress toward this very same goal.

 In FY06, IT spend as a percent of
revenue was 1.39% resulting in more than $1.4 billion in benefits. Today we’re
at 40% maintenance and 60% innovation by budget – and even more by headcount – and
we’re whittling away at the final 10 percent. We’re really interested in
hearing your ideas for getting to 30/70. Feel free to comment on this post, drop
by IdeaStorm or post on your own blog
and tag it Dell30/70 so we can find you.

HP wants to reduce
its data centers from 85 to 6
. We rationalized 30 data centers to 2 (and plan
on adding centers in the coming years). It was a painful exercise well worth
hundreds of millions of dollars in savings and the wholesale elimination of complexity
and duplication that just seems crazy in retrospect. Six strategies implemented
over the course of a few years generated most of the benefits. Here’s what we

1.       Consolidated
our business processes and applications.
For each major business process we designed ONE global process. Only
then could we select the best applications to support each process and host
them in single locations. For instance, we have global HR processes and
applications that let us manage HR for 85,000 employees from Round Rock. We
used to manage in 86 countries. We created one, global e-commerce instance,
and now all 86 versions of are on
a single global application.

2.       Migrated to fewer operating systems. At one time we literally had dozens of operating systems at Dell,
including various flavors of Unix, Solaris, Windows NT, 2000 and 2003. And,
there was no standard desktop image. Today, we have three server images (Linux,
Windows, Netware), a single corporate image for desktops and
IT managers who actually get some sleep at night.  

3.       Consolidated
our servers
. If we had not rationalized our data centers
we’d probably have more than 50,000 servers and hundreds of people managing
them today! By standardizing around fewer operating systems and applications,
we’re able to run the business with only 20,000 Dell servers and only about a dozen
people dedicated to systems management. But we’re not resting on our laurels in
server consolidation. Now we’re
rolling out virtualization as we retire old servers
. We believe we can
virtualize around 50 percent of our entire IT infrastructure in the next couple
of years, and even more as we build new applications designed from the ground
up for virtual servers. That will get us even closer to 30/70.  

4.       Rationalized
As with most of our customers, keeping
storage in control is a challenge. At one time,
we had every kind of
storage known to man. We moved to a Storage Area Network (SAN) to improve
performance and stability, as well as reduce complexity and cost. It is working.
Dell’s SAN products team has done a great job at driving down the price of
entry and implementation cost for SANs. Today, we’re almost at 7 petabytes or 7
quadrillion bytes which is one of the largest commercial SANs in

our data.
Years ago, practically every Dell team had multiple
“data marts” that included Excel spreadsheets, Access and SQL Server, and large
Oracle databases (and everything in between). Our vision was ambitious: One
data warehouse for all of Dell’s mission critical data. The project started
almost 10 years ago. Today we have one data warehouse. Here’s a good example: When
you shop on 100 percent of customers’ clicks are captured in this
database. Our team recently built a mashup – using
click data, IP addresses and Google Earth.

new data centers and closed old ones.
After some
pain, we learned that it was often a lot easier to build greenfield data centers vs. expanding
existing ones. It’s also the only way we could ensure proper facilities for
emerging technology so that’s what we did. So a few years ago we took 30 data
centers down to just two, and now we’re leveraging talent across the company to
build a couple more that will be among the most advanced and energy-efficient data
centers on the planet.

We learned a lot from these six initiatives, and we are more than happy to
“open source” our experience to help other companies including Mr. Hurd and HP move
toward 30/70. Here are three lessons learned I’ll share now:


  • Don’t starve your IT infrastructure because it will
    catch up with you
    . Under-investing in IT is sort of like putting cheap gas in the
    car and saving money by not changing fluids or doing preventive
    maintenance. The car runs great for a while, but at some point, you’re going
    to find yourself stuck on the side of the road. Similarly, with time some
    of our old “state of the art” data centers couldn’t scale with our IT and
    business requirements. Think ahead. Only build data centers that “expand”


  • IT innovation needs to stay close to the business. We learned this the hard way. We set up a development center in India
    manned with some of the best software architects and programmers in the
    world. But organizationally they were not well connected to the business
    and its leaders. Today we keep mission-critical application development
    really close to the business   and our development teams, whether they
    are in India, Brazil, China,
    Europe or the U.S.,
    report directly into the business process they support.


  • Distribute and mitigate risk.  This is an important issue for us and all
    of our customers. Our approach is rather straight-forward: On a quarterly
    basis we purposefully “break” our mission-critical application and make
    sure our back-up systems work. One thing we discovered by doing this is
    that we were overly dependent on tape. While tape is adequate for some
    applications, it’s too slow to back-up data for others. 

30/70 is
something we talk about at Dell, a lot. In the coming weeks, my colleagues and
I will share more ideas for achieving 30/70. In the meantime, we’d love to hear
from anyone out there in the IT community who has something to contribute. How
are you stripping cost and complexity from IT so you can focus more on

About the Author: Glenn Keels